HHWT (H 18-21)
Household weight
For directions on reading the variable description see Data
Dictionary Introduction.
Availability:
|
1850
|
1860
|
1870
|
1880
|
1900
|
1910
|
1920
|
1940
|
1950
|
1960
|
1970
|
1980
|
1990
|
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
X
|
Universe:
All households and group quarters.
Description/Comparability:
HHWT describes how many households in the U.S. population in a census
year are represented by a given household in the IPUMS. It must be used
to yield accurate statistics from 1940, 1950, and 1990 household-record
variables. The application of weights is discussed further in Chapter
1, "Introduction," and Chapter
2, "Sample Designs."
To obtain nationally representative statistics, HHWT should always be
used for household-level analyses in 1940, 1950, and 1990. For other years
the use of the weight is optional; as flat samples, every cases in the
IPUMS represents the same number of households in the population (approximately
equal to the sample density). The weights have been adjusted to reflect
the exact ratio of persons in the population to persons in the sample.
Because these values are not whole numbers, the appropriate number of cases
are randomly assigned to each adjacent whole value (for example, if one
case represents 100.25 households, 75% of the cases are assigned 100 and
25% are assigned 101).
The following chart indicates the number of households each case represents
in each sample (besides 1940, 1950, and 1990):
| Sample |
Weight |
Sample |
Weight |
| 1850 |
101.15 |
1970 Form 1 State |
100.13 |
| 1880 |
99.74 |
1970 Form 2 State |
100.17 |
| 1900 |
758.94 |
1970 Form 1 Metro |
100.14 |
| 1910 |
251.89 |
1970 Form 2 Metro |
100.17 |
| 1920 |
202.86 |
1980 Metro |
99.92 |
| 1960 |
99.62 |
1980 State |
20.01 |
In addition, in 1940, 1950, and 1990, the original weights in the PUMS
have been adjusted slightly to reflect the final population totals. The
1940 values were increased 1.0%, the 1950 values decreased 1.0%, and the
1990 values increased 1.7%.
1860 and 1870 are flat samples, except that households containing any
black person were sampled at twice the density of other households. Thus,
in the preliminary sample, households without blacks receive a weight of
500, whereas households with blacks receive a weight of 250.
Another weight variable, SELFWTHH,
is available in 1940 and 1950. It identifies a flat subsample in which
every case is weighted equally (i.e., no weighting is necessary). The price
of this convenience is a loss of cases.
Codes: See text above. |
